Putonghua Baby 's Two - Word Group

Phonological acquisition is an important topic in psycholinguistics. Connectionism emphasizes the interactions among different linguistic dimensions, and proposes that phonological acquisition is highly related to ambient input. Neutral tone is the unstressed syllable in Mandarin. It contains both tone and lexical stress, hence calls for multiple decoding. There is no research on the acquisition of neutral tone by Mandarin-learning infants. In the present study, based on the Connectionism, I studied how Mandarin-learning children acquire neutral tone in the first two years by investigating both production and perception.The developmental trajectory of infants’ vowels and consonants perception supports the Perceptual Reorganizaiton(PR) hypothesis. It is still unclear whether the acquisition of lexical tones fits into the PR hypothesis. With regard to lexical stress, do tone-language learning infants and stress-language learning infants show a different developmental course? To this end, Dutch(a typical stress language) was introduced as the control grouop, and cross-linguistic perception experiments were conducted. As speech perception is highly influenced by input, I also studied neutral tones in Mandarin infants’ production and Infant Directed Speech(IDS).For the analysis of IDS, 4 infant-caretakers dyads were selected from the CASS_CHILD corpus, and the infants were around one year old. According to the distributional analysis, there are three non-grammatical neutral tones, including reduplication(RE), word with suffix(SF), and irregular neutral tone(IR). The results show that(i) 30.46% of the neutral tones are non-grammatical in IDS;(ii) RE is the most frequent type in IDS, yet they tend to be misproduced as canonical tone, which is influenced by tone combinations and lexical meanings;(iii) SF and IR are produced more accurately than RE.For children’s production, recordings of toddlers at 18- and 24-month-old in the corpus of “CASS_CHILD_WORD” were analyzed. The results demonstrate that(i) 18-month-old toddlers have acquired the pitch pattern of neural tone but not the duration pattern;(ii) toddlers have acquired both the pitch and the duration pattern by 24 month olds.Since adults’ speech perception is the ultimate goal of child language development, I first tested adults’ perception of lexical tones as a baseline, aiming to understand how different phonetic cues contribute to neutral tone perception. Dutch listeners, whose native language is a typical stress language, were tested as a control group, and I investigated whether the duration and pitch information of neutral tone could trigger the perception of unstressed syllable. The results show that(i) Both Mandarin- and Dutch-listeners can identify netural tone as unstressed syllable;(ii) the Mandarin listeners are more sensitive to pitch variation whereas the Dutch listeners are more sensitive to duration variation when identifying the neutral tone;(iii) When we manipulated either duration or pitch alone, the manipulation cannot activate the recognition of neutral tone.Based on the results of production and adults’ perception, the irregular neutral tones were selected as stimuli for infants’ perceptual experiments. Two experiments were conducted, with the degree of phonetic difference between canonical and neutral tone being different:(1) large distance: /pan1san2/ vs./pan1san0/;(2) small distance: /pan1san4/ vs. /pan1san0/. Dutch-learning infants were tested in the Experiment 2 considering the Dutch intonation pattern.The visual fixation paradigm was adopted to test 4~6- and 10~12-month-old infants. 111 infants were tested, including 59 Mandarin-learning infants and 52 Dutch-learning infants. The results indicate that Dutch-learning infants can distinguish neutral tone from canonical tone at 4~6 months, but they lost the ability after 10 months, which is consistent with the PR hypothesis. The declined sensitivity might be due to the intonational pattern of Dutch.However, neither group of Mandarin-learning infants discriminated between canonical tone and neutral tone, regardless how large the acoustic distance is. As we used pseudowords, it might be that the unfamiliarity of the stimuli prevented the infants from acessing the phonological representation of neutral tone.A discrimination asymmetry was found. When infants were habituated by neutral tone, they did distinguish between the neutral tone and the canonical tone. It seems that when habituated with the neutral tone, infants learn to build up a representation of it, which helps them to identify canonical tone as not belonging to the neutral tone category.The results provide new evidence for understanding infants’ language acquistion. When neutral tone is not lexically or syntactically marked, infants younger than 12 months cannot discriminate neutral tone from canonical tone. It is likely that the representation of neutral tone is restricted to the specific lexical items, and is exemplar based. Therefore, language learning possibly a process of generalization, i.e., infants first learn the specific examplars and gradually extract the abstract phonological rules. Meanwhile, for Mandarin-learning infants, the production of neutral tone is progressive, and infants gradually update the phonetic specification of neutral tone.